Integration of multiple sensory inputs is required for robust perception and behavioral performance. Recent psychophysical studies indicate that humans combine cues according to a statistically optimal weighting scheme derived from Bayesian probability theory. When we make perceptual judgments that rely on two separate cues, we are able to take into account the reliability of each cue, even when this reliability varies randomly from trial to trial. Bayesian theory also predicts an improvement in behavioral performance when two sensory cues are present, as compared with only one cue. One particularly vital task that involves multisensory integration is the estimation of self-motion, or heading. Information from both visual (`optic flow') and vestibular cues can be useful for heading perception, yet little is known about the principles and neural substrates for cue integration. Using a sophisticated virtual reality system, we have developed a novel behavioral paradigm for studying cue integration in rhesus monkeys. Here we propose to test specific hypotheses regarding a role of the dorsal medial superior temporal area (MSTd) and ventral intraparietal area (VIP) in visual/vestibular cue integration. In aim 1, we will record from MSTd and VIP neurons while the monkey performs a heading discrimination task, based on vestibular cues alone, visual cues alone and combined presentation of both cues. We hypothesize that MSTd/VIP neurons with `congruent' visual/vestibular responses constitute a neural substrate for Bayesian cue integration, including cue re-weighting when visual reliability changes. In aim 2, we will probe for causal links between MSTd/VIP neurons and multi-sensory cue integration for heading perception by manipulating neural activity using microstimulation and reversible chemical inactivation techniques. In aim 3, we explore the functional roles of MSTd/VIP neurons having `opposite', as compared to `congruent', visual/vestibular responses. We will test specific hypotheses, the most prominent of which is a potential role of neurons with opposite visual/vestibular preferences on disambiguating the components of visual motion that are due to object motion from those due to self-motion. Combined these aims will provide a fundamental breakthrough in our understanding of multisensory cortex, as well as more generally how the brain computes using probabilities. The general principles that we uncover regarding sensory integration should have wide application to many issues in systems-level neuroscience. Understanding the neural basis of multisensory integration and self-motion perception would also promote new strategies for treating spatial disorientation deficits common to many brain dysfunctions, including Alzheimer's disease. One of these deficits is an impaired ability to judge heading from optic flow, and this impairment is correlated with patients' difficulty in navigating through their surroundings. Better localization of these functions in a primate animal model would help targeting new Alzheimer's therapies to the appropriate brain regions. PUBLIC HEALTH RELEVANCE Understanding the neural basis of multisensory integration and self-motion perception would also promote new strategies for treating spatial disorientation deficits common to many brain dysfunctions, including Alzheimer's disease. One of these deficits is an impaired ability to judge heading from optic flow, and this impairment is correlated with patients' difficulty in navigating through their surroundings. Better localization of these functions in a primate animal model would help targeting new Alzheimer's therapies to the appropriate brain regions.